How to perform a paired t-test in R when all the values are in one column?

问题

The common format of a data frame for peforming a Paired t-test in R is a given measurement divided in two columns, so that each row indicates a same subject or individual. For example:

 > #     Before     After
    > #1      31        32
    > #2      22        34
    > #3      41        35

However, this is not always the case in which the data are presented in a data frame.

I have a data frame which structure is very common and looks like this:

 subject <- c("A1", "A2" ,"A1" ,"A3" ,"A3" ,"A2")
    value <- c(34, 43, 25, 43, 54, 22)
    group <- c("before", "after", "after", "after", "before", "before")

    mydata <- data.frame(subject, value, group)

    #  subject value  group
    #1      A1    34 before
    #2      A2    43  after
    #3      A1    25  after
    #4      A3    43  after
    #5      A3    54 before
    #6      A2    22 before

So, based on this data frame how can I perform a two-sided paired t-test in R?

回答1:

What about some R base function to have data from long to wide format:

# reshape from long to wide
mydata_wide <- reshape(mydata, idvar = "subject", timevar = "group", direction = "wide")

# rename columns
colnames(mydata_wide)[2] <-"before"
colnames(mydata_wide)[3] <-"after"

mydata_wide
  subject before after
1      A1     34    25
2      A2     22    43
4      A3     54    43

# t-test
 t.test(mydata_wide$before,
       mydata_wide$after,
       paired=TRUE,
       conf.level=0.95)

来源：https://stackoverflow.com/questions/52461376/how-to-perform-a-paired-t-test-in-r-when-all-the-values-are-in-one-column